Published in "European Journal of Public Health 30(3): 552–557, 2020" which should be cited to refer to this work.

Cervical cancer (over)screening in and : trends and social inequalities

Vincent De Prez1, Vladimir Jolidon 2, Barbara Willems1, Ste´ phane Cullati2,3, Claudine Burton-Jeangros2, Piet Bracke1

1 Department of Sociology, Ghent University, Ghent, Belgium 2 Institute of Sociological Research, Geneva School of Social Sciences, University of Geneva, Geneva, Switzerland 3 Population Health Laboratory, Department of Community Health, Faculty of Science and Medicine, University of Fribourg, Fribourg, Switzerland

Correspondence: Vincent De Prez, Department of Sociology, Ghent University, Korte Meer 5, 9000 Ghent, Belgium, Tel: þ32 (0) 9 264 68 12, Fax: þ32 (0) 9 264 69 75, e-mail: [email protected]

Background: Cervical cancer screening (CCS) by means of Pap smears has led to a decrease in cervical cancer incidence and mortality. In the absence of organized programmes, CCS is opportunistic in Belgium and Switzerland. This might result in a high level of CCS overuse, as screening practices do not conform to the recommended 3-yearly screening interval and the target age-ranges (Belgium: 25–64, Switzerland: 20–70). This study aimed to assess trends in CCS uptake and overuse in Belgium and Switzerland and their social determinants, in the light of reimbursement initiatives, which were implemented in both countries. Methods: Data from five waves of the Belgian Health Interview Survey (1997–2013) (N¼11 141) and Swiss Health Interview Survey (1992– 2012) (N¼32 696) were used. We performed Poisson regressions to estimate adjusted prevalence ratios (APR), controlled for socio-economic and socio-demographic characteristics and health status. CCS overuse was opera- tionalized as screening more than once every 3 years and screening above recommended age-range. Results: CCS uptake remained relatively stable over time, with a mean coverage of 70.9% in Belgium and 73.1% in Switzerland. Educational and income gradients were found in both countries. Concerning CCS overuse, women above screening-eligible age showed consistently high screening rates, but screening within the past year declined significantly in both countries, matching the temporal implementation of the reimbursement initiatives. Conclusions: Although no increase in CCS coverage could be established, CCS has become more efficient in both countries as Pap smear overuse at the population level has declined after the implementation of reimburse- ment measures tackling CCS overuse......

Introduction level,4,19 heterogeneous quality,20 psychological distress21 and false positives due to over-diagnosis.15 ncidence and mortality in cervical cancer have declined since the A decline in CCS overuse has been reported in observational and 1–3 I1970s in high-income Western-European countries. These intervention studies in some European countries,16,22,23 both within trends have been linked to the introduction of cervical cancer and outside of the recommended age-range. In order to further http://doc.rero.ch screening (CCS) by means of Pap smears.4–6 Consequently, the picture trends in CCS overuse and to identify potential pathways has recommended 25- to 64-year-old women to to explain a possible reduction in overuse of preventive services, 4 undergo CCS screening on a 3-yearly interval. In organized systems, more research is needed.16 This research offers foothold in this la- eligible women are periodically invited to undergo screening, where- cuna by (i) making a cross-country inquiry into the evolution as in opportunistic systems, the initiative to comply to the screening of CCS uptake and overuse, and (ii) linking the time trends to recommendations is left to physicians and individual women.7 If government initiatives. This is facilitated by comparing Belgium women undergo Pap smears more than once within the recom- and Switzerland. These Western-European countries both rely on mended 3-yearly interval or outside the 25- to 64-year-old target opportunistic screening,4,7,24 and have implemented a similar initia- age-range, this could be considered CCS overuse, which has been tive to limit overuse at national level. However, these countries have found to be more likely in opportunistic systems.4 In the first form different preventive healthcare systems. The Swiss healthcare system of CCS overuse, women within screening-eligible age undergo CCS is more liberal and fragmented between regions than the Belgian 8–13 too frequently. In the second form, past research had identified system which is more uniformly coordinated by the federal govern- 14,15 both women below the recommended age and women above the ment. The Swiss system also relies much more on out-of-pocket recommended age16–18 for screening. CCS overuse entails several expenditure compared to the Belgian one when considering prevent- detrimental aspects, such as low cost-effectiveness at population ive care.

1 CCS overuse has been observed in Belgium since the 1990s,9,25 respondent was screened within the last year (yes ¼ 1, no ¼ 0), and has declined from a 3-yearly average of 1.88 smears per women among women who have been screened within the last 3 years. This in 200026—1.18 smears in 2010–12.27 In Switzerland, it was esti- dependent variable operationalizes our second definition of screen- mated that a screening coverage of 70% is achieved by the 1–1.2 ing overuse based on the screening frequency and understands over- million Pap smears that are taken annually.28,29 Nevertheless, use as screening more than once within the recommended 3-yearly 520 000 smears would be enough to cover 100% of the eligible time frame.30,35 Hence, if women answered ‘yes’ to this question, population.28,29 In both countries, similar policy incentives affecting they were classified in the overuse category. Among the screening CCS overuse were implemented. The Swiss government has, accord- women, if the proportion of women who stated that they were ing to the 1996 Swiss health insurance law, reimbursed one CCS screened within the last year was >33.33% (under the assumption every 3 years for 18- to 69-year-old women.30 More than a decade of an even distribution over the 3 years) we used this as a proxy later, in Belgium, reimbursement was limited to one Pap smear indicator of overuse at the population level (for a detailed outline of every 2 years since 2009, and to once every 3 years since 2013.27,31 this operationalization, see Supplementary material 2). We expect to establish a corresponding impact of these policy meas- ures in both countries. Furthermore, attention is directed towards Independent variables inequalities in screening uptake. Previous research in Belgium and Indicators of interest and covariates were selected based on their Switzerland has found that adherence to CCS guidelines is related to 7,34 the individual’s education, financial situation and socio- potential association with CCS, as indicated by previous research. demographic profile.4,7,32 Women with the lowest probability of We included the following indicators of interest: education, monthly partaking in CCS generally have a lower educational attainment, a household income, employment status, age, nationality and urban- lower household income, an older age, no partner and live in a rural ization. To control for known associations with CCS, we took into area.4 These disparities are shown to be persisting over time.7,33,34 account the following covariates: having a partner, self-rated health, smoking, body mass index (BMI) and having visited a physician Accordingly, we expect social inequalities to be consistent—both for 7,36–38 CCS uptake and for CCS overuse. within the last 12 months (for an operationalization of these In summary, this study aims at assessing the prevalence and trends variables, see Supplementary material 3). in CCS uptake and CCS overuse in Belgium and Switzerland, in the light of governmental reimbursement initiatives. Furthermore, it Statistical analyses builds on previous research concerning the social gradient in prevent- Prevalence rates of CCS uptake and CCS overuse were reported ive health behaviour by examining whether manifestations of socio- using weighted proportions and were stratified by year. Socio- economic inequalities are comparable in Belgium and Switzerland. economic inequalities in CCS uptake and CCS overuse were examined by estimating adjusted prevalence ratios (APR) and their corresponding 95% confidence intervals (CI) using Poisson regres- Methods sion models. Robust variance estimators were used to make sure that Sample the errors were not heteroskedastic. Models were adjusted for the above-mentioned independent variables. Survey wave was included Data used for this study were obtained from the Belgian Health in our models as a continuous variable to take fluctuations over time Interview Survey (BHIS) and the Swiss Health Interview Survey into account. A detection for multicollinearity among socio- (SHIS). Both cross-sectional surveys contain five waves: 1997, economic variables using variance inflation factors showed no 2001, 2004, 2013 for Belgium; and 1992, 1997, 2002, 2007, 2012 issues. Employment status was not included in the model for the for Switzerland. For the BHIS, a random multistage probability 65–75 years old due to too little variation across categories. Trends sample was drawn from the National Register to select households were examined for both definitions of CCS overuse. Analyses were and their members. Data were collected through face-to-face inter- weighted for survey sampling and non-participation bias and were views and a self-administered questionnaire. For the SHIS, respond- conducted with SPSS 22 and STATA 15. To assess the robustness of ents were invited for participation based on a stratified random our findings, we replicated the analysis testing different coding selection. Data were collected through telephone interviews and a schemes for variables containing multiple categories (education self-administered questionnaire. In order to take the national con- with three or five categories, employment in two or three categories text into consideration, sample age-ranges were defined according to and included in the model with women above screening-eligible age, http://doc.rero.ch the country-specific screening recommendation guidelines. For age and income as continuous and categorical variables). We con- Belgium, 9314 women within recommended age (25–64) and 1827 ducted an analysis on the CCS uptake and CCS overuse models with women above recommended age (65–75) were examined. For samples limited to women aged 28–64 and 68–75 in Belgium and Switzerland, 30 773 women within recommended age (20–70) and 23–70 and 74–75 in Switzerland to check for misclassification due to 1923 women above recommended age (71–75) were examined. threshold effects. The interpretation of the results remained the More information on respondent selection criteria is available (see same (results in Supplementary material 4). Supplementary material 1). Results Dependent variables Pap smear uptake according to participants’ In both national surveys, participants were asked how long ago they had had a CCS. Answers to this question were computed into two characteristics dependent variables. At the individual level, the first dependent Table 1 shows that screening coverage in Belgium was higher among variable, up-to-date CCS, measures whether the respondent was women with higher education (79.0%) compared to women with screened within the last 3 years (yes ¼ 1, no ¼ 0). This dependent lower (54.3%) or intermediate (69.1%) education. In Switzerland, variable also corresponds with our first operationalization of over- lower educated women (57.8%) showed remarkably lower screening use, i.e. overuse as screening uptake above the screening- rates compared to women with an intermediate (72.8%) and a recommended age-range. The outlying age categories were 65–75 higher level of educational attainment (76.2). The income gradient for Belgium and 71–75 for Switzerland. Women below screening- in CCS uptake was more pronounced in Belgium compared to eligible age were not taken into account in this research due to Switzerland. In Belgium, women with the lowest and highest income insufficient sample sizes for this age-group. At the population level, showed a difference in screening coverage of about 25% points (first the second dependent variable, CCS overuse, measures whether the quintile: 54.2%; fifth quintile 79.0%), whereas this difference

2 Table 1 Weighted proportionsa of CCS in the past 3 years among 11 141 women aged 25–75 from the BHIS waves and 32 696 women aged 20–75 from the SHIS waves, according to women’s characteristics

Belgium Switzerland

Survey wave 1997–2013 1997 2001 2004 2008 2013 1992–2012 1992 1997 2002 2007 2012 N 11 141 2534 2414 2330 1913 1950 32 696 5469 3982 7648 7066 8531 % %%%%%P-valuesb % %%%%%P-valuesb

Education <0.001 <0.001 Lower 54.3 57.0 54.5 50.9 52.9 56.1 57.8 56.4 61.7 59.0 52.9 57.3 Intermediate 69.1 70.8 66.6 72.0 70.0 66.0 72.8 74.0 78.0 72.5 70.3 70.6 Higher 79.0 77.4 75.7 80.1 80.0 81.0 76.2 76.7 80.6 77.6 73.5 76.6 Household income <0.001 <0.001 First quintile 54.2 54.1 54.4 51.1 53.0 58.5 63.3 60.1 66.6 64.2 66.5 62.0 Second quintile 60.3 61.0 57.2 59.3 56.9 67.8 71.4 77.8 72.0 71.4 68.4 69.3 Third quintile 66.1 71.8 61.5 65.0 63.6 67.3 71.4 65.9 79.4 71.3 67.7 72.7 Fourth quintile 73.5 74.4 71.4 75.3 75.0 72.0 75.1 78.4 77.6 74.4 70.8 74.4 Fifth quintile 79.0 76.1 75.0 79.1 83.4 80.2 77.8 80.7 83.5 75.7 75.2 77.6

a: Crude rates, unadjusted for other indicators of interest and confounders. b: Weighted Pearson chi-square test for the entire studied period. P-values are lower than 0.001, which indicate that the categories of the variables are significantly different from each other.

Figure 1 CCS uptake, weighted prevalence among eligible women in Belgium (BHIS waves 1997–2013) and Switzerland (SHIS waves 1992– 2012). Screening uptake within the last 3 years, APR for time: Belgium ¼ 95% CI 1.00–1.02, and Switzerland ¼ 95% CI 0.98–0.99 (see table 2). Screening uptake within the last year, APR for time: Belgium ¼ 95% CI 0.94–0.97, and Switzerland ¼ 95% CI 0.90–0.93 (see table 3). APR’s adjusted for education, household income, employment status, age, nationality, urbanization, having a partner, self-rated health, smoking, BMI and physician visit in the last 12 months. (a) Women who had a Pap smear within the last year—33.33%

consisted of 15% points in the Swiss sample (first quintile: 63.3%; between 1997 and 2012. CCS overuse declined significantly by 4% http://doc.rero.ch fifth quintile: 77.8%). (APR ¼ 0.96, 95% CI ¼ 0.94–0.97) in Belgium, and 9% (APR ¼ 0.91, 95% CI ¼ 0.90–0.93) on average by survey wave in Trends in CCS uptake and CCS overuse Switzerland (table 3). Thirdly, the mean proportion of non-eligible women with an up-to-date Pap smear was 38.0% in Belgium and Firstly, CCS uptake among screening-eligible women has remained 38.6% in Switzerland, so screening among women above the rec- stable over the five survey waves in Belgium, with a mean screening ommended age-group was substantial (see Supplementary material coverage of 70.9% (figure 1, Supplementary material 5). No increase ¼ 6). These proportions did not change significantly over the studied in screening coverage could be established (APR 1.01, 95% period (see Supplementary material 7). CI ¼ 1.00–1.02, table 2). The mean screening coverage in Switzerland was slightly higher at 73.1%. Changes over time were significant in Switzerland, with a slight decrease over the studied Inequalities and trends in CCS and CCS overuse period (APR ¼ 0.99, 95% CI ¼ 0.98–0.99, table 2). Secondly, CCS overuse among screening-eligible women remained stable in Cervical cancer screening Belgium between 1997 and 2008, with about one-third of the In both countries, women with a lower educational level, lower house- women with an up-to-date Pap smear performing CCS overuse at hold income, who were older and who were foreign had a lower prob- population level (figure 1, Supplementary material 5). In the subse- ability to have had a Pap smear within the past 3 years (table 2). In quent period, between 2008 and 2013, only about one-fifth of the Switzerland, women living in rural areas had a lower screening preva- women with an up-to-date Pap smear performed CCS overuse at lence, and in Belgium, CCS was lower among non-employed women. population level. In Switzerland, CCS overuse was substantial in As shown by the time-interaction term, prevalence for employment 1992, where one-fifth of the women with an up-to-date Pap smear status fluctuated significantly over the five waves in Belgium performed overuse at population level. This proportion declined (P < 0.05), however no clear trend was observed (Supplementary strongly in the subsequent years, and almost no overuse was present material 8). In Switzerland, the time-interaction term indicated that

3 Table 2 Weighted and APR for up-to-date Pap smeara, among 9314 women aged 25–64 from the BHIS and 30 773 women aged 20–70 from the SHIS, according to women’s characteristics, with trends over time

Belgium P-values for Switzerland P-values for trend over trend over Up-to-date Pap smeara (yes vs. no)1997–2013b Up-to-date Pap smeara (yes vs. no) 1992–2012b

APR 95% CI APR 95% CI

Socio-economic characteristics Education (ref: lower) 0.200 0.513 Intermediate 1.09 1.04–1.15 1.14 1.10–1.17 Higher 1.17 1.11–1.23 1.13 1.09–1.17 Household income (ref: first quintile) 0.541 0.191 Second quintile 1.14 1.06–1.22 1.05 1.02–1.08 Third quintile 1.15 1.07–1.23 1.07 1.04–1.10 Fourth quintile 1.16 1.08–1.24 1.09 1.06–1.12 Fifth quintile 1.19 1.11–1.27 1.09 1.06–1.13 Employment status (ref: employed) 0.024 0.589 Unemployed 1.05 0.98–1.12 0.97 0.91–1.04 Non-employed 0.94 0.90–0.99 1.01 0.99–1.03 Socio-demographic characteristics Age (ref BE: 25–34; ref CH: 20–29) 0.191 <0.001 BE: 35–44; CH: 30–39 1.01 0.97–1.05 1.15 1.12–1.18 BE: 45–54; CH: 40–49 1.01 0.97–1.06 1.15 1.12–1.18 BE: 55–64; CH: 50–59 0.90 0.85–0.96 1.08 1.04–1.11 CH: 60–70 0.87 0.84–0.90 Nationality (ref: national) 0.429 0.579 Foreign 0.87 0.80–0.93 0.90 0.87–0.92 Urbanization (ref: urban) 0.575 0.992 Rural 1.00 0.96–1.04 0.95 0.93–0.97 Survey wave 1.01 1.00–1.02 0.99 0.98–0.99 Constant 0.44 0.39–0.49 0.42 0.40–0.45

Variables used for adjustment: having a partner, self-rated health, smoking, BMI and physician visit in the last 12 months. a: Up-to-date Pap smear: Pap smear within the last 3 years. b: P-values for time trends were estimated as follows: for each predictor, we estimated separately one multivariate model including all predictors plus the interaction term between the predictor and the wave. We reported only the P-values. APR, adjusted prevalence ratios.

the prevalence for age fluctuated significantly over the five waves Belgium in the period between 2008 and 2013. These results match (P < 0.001), with the gap between women aged 60–70 and those the findings from a study on individual health insurance data in between 20 and 29 diminishing over time (Supplementary material 8). Belgium over the same period,27 which identified a decline in CCS overuse since 2009. In Belgium, the shift was preceded by the limited-reimbursement initiative implemented in 2009.27 In CCS overuse Switzerland, the observed phenomenon was preceded by the 1996 For CCS overuse in terms of screening frequency, in both countries, Swiss health insurance law, which initiated a regime that reimburses women with an older age and living in a rural area had a lower one Pap smear every 3 years for 18- to 69-year-old women.30 Hence, likelihood of being screened within the last year (table 3). No edu- as part of the shifts from a high proportion of women undergoing cational inequalities and negligible income inequalities were found. CCS within the last year to the second year might be attributed to

http://doc.rero.ch In Switzerland, foreign women had a higher prevalence of being the reimbursement initiatives in both countries, our findings sup- screened within the last year. For CCS overuse in terms of port the idea that reimbursement initiatives can effectively alter CCS screening-recommended age, no inequalities were found among overuse. We recommend further explorations on this relation. women above screening-eligible age (Supplementary material 7). Furthermore, our findings are in line with the general declining Only in Switzerland, women living in a rural area were less often trend of CCS overuse that has been found in longitudinal data in screened. No significant fluctuations over time could be established. six observational studies.16 In spite of the declining trend in CCS overuse at population level, screening uptake among women above screening-eligible age showed no decline over the studied period. Discussion Screening uptake above the recommended age-range remains sub- In this study, the evolutions of CCS uptake and overuse were com- stantial, with nearly half of the 65- to 75-year-old women perform- pared between Belgium and Switzerland, which both have oppor- ing screening in Belgium, and two-fifths of the 71- to 75-year-old tunistic screening strategies. Trends over a 20-year period were women doing so in Switzerland. assessed using five waves of the BHIS and SHIS. Pap smear uptake Next, we found educational and income gradients in CCS uptake, with higher uptake among higher educated and wealthier women. This remained relatively stable in both countries over the studied period, 4,7,9 with a slight decrease in CCS uptake in Switzerland. CCS overuse at finding is in line with previous studies. Possible explanatory mech- anisms for the higher screening uptake of highly educated women are: population level declined remarkably in both countries, but not 39 simultaneously. An enquiry into the 3-yearly distribution of up- a higher health literacy and a more future-oriented attitude. These to-date Pap smears has indicated that CCS overuse at population factors promote a more preventive healthcare-oriented decision-mak- level has shifted from screening within the last year to screening ing. Furthermore, in our study, no socio-economic gradient in Pap between 1 year ago and 2 years ago. In Switzerland, overuse at smear overuse could be established, nor in terms of screening fre- population level declined tremendously between 1992 and 1997 quency, neither in terms of screening above eligible age. No changes and remained stable thereafter. A similar shift was observed in over time in socio-economic inequalities were found.

4 Table 3 Weighted and APR for CCS overusea in the past year, among 6676 women aged 25–64 from the BHIS and 22 099 women aged 20–70 from the SHIS, according to women’s characteristics, with trends over time

Belgium P-values for Switzerland P-values for trend over trend over Overusea 1997–2013b Overusea 1992–2012b

APR 95% CI APR 95% CI

Socio-economic characteristics Education (ref: lower) 0.674 0.728 Intermediate 1.03 0.96–1.10 1.00 0.94–1.07 Higher 1.06 0.99–1.14 1.02 0.95–1.11 Household income (ref: first quintile) 0.831 0.294 Second quintile 1.10 1.00–1.20 1.00 0.94–1.07 Third quintile 1.05 0.95–1.14 1.01 0.95–1.08 Fourth quintile 1.05 0.96–1.15 1.05 0.99–1.12 Fifth quintile 1.07 0.98–1.18 1.07 0.99–1.15 Employment status (ref: employed) 0.899 0.352 Unemployed 1.09 0.99–1.19 0.89 0.75–1.05 Non-employed 1.00 0.93–1.08 1.03 0.98–1.08 Socio-demographic characteristics Age (ref BE: 25–34; ref CH: 20–29) 0.230 0.237 BE: 35–44; CH: 30–39 0.93 0.88–0.99 0.96 0.91–1.02 BE: 45–54; CH: 40–49 0.95 0.89–1.01 0.95 0.90–1.01 BE: 55–64; CH: 50–59 0.82 0.75–0.90 0.91 0.85–0.97 CH: 60–70 0.82 0.75–0.89 Nationality (ref: national) 0.232 0.180 Foreign 1.03 0.94–1.13 1.06 1.00–1.13 Urbanization (ref: urban) 0.070 0.936 Rural 0.92 0.87–0.97 0.95 0.91–1.00 Survey wave 0.96 0.94–0.97 0.91 0.90–0.93 Constant 0.42 0.36–0.51 0.15 0.13–0.18

Variables used for adjustment: having a partner, self-rated health, smoking, BMI and physician visit in the last 12 months. a: Overuse among screeners (<1 vs. 1–3). b: P-values for time trends were estimated as follows: for each predictor, we estimated separately one multivariate model including all predictors plus the interaction term between the predictor and the wave. We reported only the P-values. APR, adjusted prevalence ratios.

Several strengths and weaknesses of this study must be acknowl- research due to a lack of respondents in that age-group in the SHIS. edged. Firstly, a strength of the research lies in having operational- It is recommended for future research to include these women at the ized CCS overuse in a 2-fold manner. Apart from screening outside bottom of the age-spectrum in the assessment of CCS overuse. of the recommended age-range, which had already been established In conclusion, this population-based comparative study points at in the literature as an indicator of overuse,14,17 this study has also the relevance of restricting reimbursement conditions as a means to addressed CCS overuse in terms of screening frequency at the popu- reduce Pap smear overuse. Whereas CCS uptake remained relatively lation level, and combined both approaches. This 2-fold operation- stable over the studied period in Belgium and Switzerland, CCS alization of overuse is an innovative approach in survey-based overuse in terms of screening frequency at the population level research, as it has only been previously included in research where has shown a declining trend. On the one hand, screening has be- data on individual health records were available.9,11 Secondly, this come more efficient as the amount of Pap smears taken has become

http://doc.rero.ch research has linked CCS overuse-related policy initiatives to trends less concentrated in a select group of women. At the same time, the in CCS uptake. Nevertheless, in the absence of a direct indicator, we overall screening coverage has not increased. Socio-economic were not able to measure the impact of the policy changes directly. inequalities in CCS uptake remain relevant in the Belgian and Furthermore, as other institutional factors (prevention policies and Swiss opportunistic screening regimes. For CCS overuse, no socio- healthcare system features) might have played their role in the economic inequalities were identified. decreasing CCS overuse, the impact of reimbursement initiatives must be interpreted with caution. We recommend future studies to examine how the shift in CCS overuse at the population level Funding has evolved in the subsequent years, beyond the available survey waves for this research. Thirdly, the estimation of CCS overuse This work was supported by a grant of Fonds Wetenschappelijk in terms of frequency is an approach at the aggregated level. Onderzoek (grant number FWOOPR2018005701) and Consequently, our results do not provide information on overuse Schweizerischer Nationalfonds zur Fo¨rderung der Wissenschaftlichen at the individual level. Fourthly, the choice of predictors for CCS was Forschung (grant number 176115). based on the availability of comparable questions and indicators in the BHIS and SHIS datasets. Crucial information on incentives for Pap smear uptake is missing, as it was not possible to include indicators, Data availability such as having symptoms of cervical lesions, or having had sexual This study used the data from the Belgium Health Interview Survey intercourse. Fifthly, in the BHIS and SHIS, the participation in CCS and the Swiss Health Interview Survey. The Belgian data are avail- is self-reported information, which might be subject to recall bias and able for researchers, but according to the Belgian legislation an au- an overestimation of uptake, as people tend to underestimate the time thorization has to be obtained from the Belgian Commission for the since their previous participation in preventive services. Lastly, women Protection of Privacy. The Swiss data are available for fee (1600 below screening-eligible age were not taken into account in this Swiss Francs, plus 7.7% tax) and users must request permission

5 from the Swiss Federal Statistical Office ([email protected]). Data 17 Guo F, Hirth JM, Berenson AB. Cervical cancer screening among women 70 years must be destroyed after 5 years. of age in the United States—a referral problem or patient choice. Prev Med 2015;81: 427–32. Conflicts of interest: None declared. 18 Salloum RG, Kohler RE, Jensen GA, et al. US Preventive Services Task Force recommendations and cancer screening among female Medicare beneficiaries. Key points J Womens Health (Larchmt) 2014;23:211–7. 19 Elfstro¨m KM, Arnheim-Dahlstro¨m L, von Karsa L, Dillner J. Cervical cancer • Restricting reimbursement in Belgium and Switzerland screening in : quality assurance and organisation of programmes. Eur seemed effective in reducing CCS overuse in terms of screen- J Cancer 2015;51:950–68. ing frequency, which implies an improved efficiency from a 20 Miller A. The (in) efficiency of cervical screening in Europe. Eur J Cancer 2002;38: public health perspective. 321–6. • Educational and income gradients were found in CCS uptake 21 Sawaya GF, Kulasingam S, Denberg TD, Qaseem A; Clinical Guidelines Committee in both countries, and persisted over time. This indicates that of the American College of Physicians. Cervical cancer screening in average-risk the Belgian and Swiss healthcare systems need to pay more women: best practice advice from the Clinical Guidelines Committee of the attention to health inequalities. American College of Physicians. Ann Intern Med 2015;162:851–9. • No educational or income gradient was found in CCS overuse. 22 Broach V, Day L, Barenberg B, et al. Use of electronic health record–based tools to improve appropriate use of the human papillomavirus test in adult women. J Low Genit Tract Dis 2014;18:26–30. References 23 White P. Effects of electronic health record–based interventions on cervical cancer screening in adolescents: a 1-year follow-up. J Low Genit Tract Dis 2014;18:169–73. 1 Benard VB, Thomas CC, King J, et al.; Centers for Disease Control and Prevention 24 Arbyn M, Van Oyen H. Analysis of individual health insurance data pertaining to (CDC). Vital signs: cervical cancer incidence, mortality, and screening—United Pap smears, colposcopies, biopsies and surgery on the uterine cervix (Belgium, States, 2007–2012. MMWR Morb Mortal Wkly Rep 2014;63:1004–9. 1996–2000). WIV-ISP, 2004. 2 Arbyn M, Raifu AO, Weiderpass E, et al. Trends of cervical cancer mortality in the 25 Arbyn M, Van Oyen H. Cervical cancer screening in Belgium. Eur J Cancer 2000;36: member states of the European Union. Eur J Cancer 2009;45:2640–8. 2191–7. 3 Levi F, Lucchini F, Negri E, et al. Cervical cancer mortality in young women in 26 Arbyn M, Benoy I, Simoens C, et al. Prevaccination distribution of human papil- Europe: patterns and trends. Eur J Cancer 2000;36:2266–71. lomavirus types in women attending at cervical cancer screening in Belgium. Cancer 4 Arbyn M, Rebolj M, De Kok IM, et al. The challenges of organising cervical Epidemiol Biomarkers Prev 2009;18:321–30. screening programmes in the 15 old member states of the European Union. Eur J 27 IMA-AIM. Baarmoederhalskankerscreening 2006-2012, derde IMA-rapport. 2015. Cancer 2009;45:2671–8. 28 Petignat P, Untiet S, Vassilakos P. How to improve cervical cancer screening in 5 Sasieni P, Adams J, Cuzick J. Benefit of cervical screening at different ages: evidence Switzerland? Swiss Med Wkly 2012;142:w13663. from the UK audit of screening histories. Br J Cancer 2003;89:88–93. 29 Petignat P, Bouchardy C, Sauthier P. Cervical cancer screening: current status and 6 Zappa M, Visioli C, Ciatto S, et al. Lower protection of cytological screening for perspectives. Rev Med Suisse 2006;2:1308–9. adenocarcinomas and shorter protection for younger women: the results of a case– 30 Petersen EE, Scholl B, Sarasin E, et al. Guideline zum Vorgehen bei suspektem und control study in Florence. Br J Cancer 2004;90:1784–6. positivem zytologischen Abstrich der Cervix uteri (u¨berarbeitete Fassung, Version 7 Burton-Jeangros C, Cullati S, Manor O, et al. Cervical cancer screening in 2.4). Gynakol Geburtshilfliche Rundsch 2005;45:44–55. Switzerland: cross-sectional trends (1992–2012) in social inequalities. Eur J Public 31 Onkelinx L. Koninklijk Besluit in verband met terugbetaling van cervixuitstrijkjes. Health 2017;27:167–73. Arreˆte´ Royal Concernant le Remboursement Des Frottis Cervico-Vaginaux. Belgisch 8 Almeida CM, Rodriguez MA, Skootsky S, et al. Cervical cancer screening overuse Staatsblad/Moniteur Belge 2013;22088–22089:11882–3. and underuse: patient and physician factors. Am J Manag Care 2013;19:482–9. 32 Willems B, Bracke P. The impact of regional screening policies on the diffusion of 9 Arbyn M, Fabri V, Temmerman M, Simoens C. Attendance at cervical cancer cancer screening participation in Belgium: time trends in educational inequalities in screening and use of diagnostic and therapeutic procedures on the uterine cervix Flanders and Wallonia. BMC Health Serv Res 2018;18:943. assessed from individual health insurance data (Belgium, 2002-2006). PLoS One 33 Pale`ncia L, Espelt A, Rodrı´guez-Sanz M, et al. Socio-economic inequalities in breast 2014;9:e92615. and cervical cancer screening practices in Europe: influence of the type of screening 10 Arbyn M, Quataert P, Van GH, Van HO. Cervical cancer screening in the Flemish program. Int J Epidemiol 2010;39:757–65. http://doc.rero.ch region (Belgium): measurement of the attendance rate by telephone interview. Eur J 34 Willems B, Bracke P. The education gradient in cancer screening participation: a Cancer Prev 1997;6:389–98. consistent phenomenon across Europe? Int J Public Health 2018;63:93–103. 11 Arbyn M, Simoens C, Van Oyen H, et al. Analysis of 13 million individual patient 35 Arbyn M, Anttila A, Jordan J, et al. European guidelines for quality assurance in records pertaining to Pap smears, colposcopies, biopsies and surgery on the uterine cervical cancer screening—summary document. Ann Oncol 2010;21:448–58. cervix (Belgium, 1996–2000). Prev Med 2009;48:438–43. 36 Carrasco-Garrido P, Hernandez-Barrera V, Lopez de Andres A, et al. Awareness and 12 Barbadoro P, Ricciardi A, Di Tondo E, et al. Utilization patterns of cervical cancer uptake of colorectal, breast, cervical and prostate cancer screening tests in Spain. screening in . Eur J Cancer Prev 2015;24:135–40. Eur J Public Health 2014;24:264–70. 13 Verrilli L, Winer RL, Mao C. Adherence to cervical cancer screening guidelines by 37 Tanton C, Soldan K, Beddows S, et al. High-risk human papillomavirus (HPV) gynecologists in the Pacific Northwest. J Low Genit Tract Dis 2014;18:228–34. Infection and cervical cancer prevention in Britain: evidence of differential uptake 14 Henderson JT, Harper CC, Gutin S, et al. Routine bimanual pelvic examinations: of interventions from a probability survey. Cancer Epidemiol Biomarkers Prev 2015; practices and beliefs of US obstetrician-gynecologists. Am J Obstet Gynecol 2013; 24:842–53. 208:109.e1–7. 38 Leung GM, Woo PP, Cowling BJ, et al. Who receives, benefits from and is harmed 15 Van Kerrebroeck H, Makar A. Cervical cancer screening in Belgium and overscre- by cervical and breast cancer screening among Hong Kong Chinese? J Public Health ening of adolescents. Eur J Cancer Prev 2016;25:142–8. (Oxf) 2008;30:282–92. 16 Alber JM, Brewer NT, Melvin C, et al. Reducing overuse of cervical cancer 39 Ratzan SC. Health literacy: communication for the public good. Health Promot Int screening: a systematic review. Prev Med 2018;116:51–9. 2001;16:207–14.

6